Uncertainty and sensitivity analysis in soil strata model generation for ground settlement risk evaluation

Abstract

Popular summary:
Ground settlement due to groundwater drainage during construction is important to be considered since ground settlement may cause severe building damages. The calculation of ground settlement contains several parameters with uncertainties, e.g., soil layer thickness and elevations, compression and consolidation properties of different soils. If the uncertainties in the parameters are defined, the uncertainties in ground settlement could be estimated. By addressing the most influential parameters on the ground settlement in a certain area, the uncertainties in the ground settlement could be decreased. In this thesis, a three dimensional soil layer model was generated for ground settlement risk evaluation purpose. The accurate prediction of soil layers around the construction site could help better estimation of ground settlement. The uncertainties in the soil layer model were quantified and then the soil layer model was integrated with two other models (groundwater and ground settlement model) for the ground settlement risk evaluation. The case study site was located in Motala, Sweden where a pedestrian tunnel was considered to be built.
The soil layer model was interpolated from borehole data - holes drilled into the ground for soil layer investigation. The interpolation technique used here was called kriging. The uncertainties in the soil layer model came from the interpolation process and were quantified. The risk evaluation of the ground settlement was carried out by Monte Carlo simulation which is a kind of stochastic modeling method.
The resulted soil layer model could show soil layers formation at any specified site. The quality of interpolation was largely dependent on the amount of boreholes. The soil layer model needed to be adjusted manually since sometimes the interpolated soil layers overlapped each other. The ground settlement risk evaluation results illustrated the risk areas for building damage where the calculated ground settlement exceeded critical values. The most influential parameters on ground settlement were found varied in different places. More efforts and resources could be spent on these parameters to decrease the unacceptable risks. The kriging method used here was a sophisticated method and there is room for improvements.

Abstract

Scientific summary:
Ground settlement due to groundwater drainage during construction is important to be considered since ground settlement may cause severe building damages. The calculation of ground settlement contained several parameters with different magnitude of uncertainties. Thus a risk evaluation of ground settlement is necessary. The aim of this thesis was first to build a soil strata model for ground settlement risk evaluation purpose. Second was to carry out the uncertainty and sensitivity analysis of the soil strata model. Third was to carry out the ground settlement risk evaluation by integrating soil strata model and two other models, with defined uncertainties of each model. The case study site was located in Motala, Sweden with area about 0.39 km2.
The soil strata model was generated by utilizing kriging interpolation. The continuous elevations of each soil layer in the soil strata were interpolated from boreholes and then all the soil layers were combined to create a “layer-cake model”. The uncertainty in kriging was quantified by prediction standard error. By utilizing Monte Carlo simulation, the stochastic representation of the soil strata was created and the uncertainty and sensitivity analysis of the soil strata model was carried out. The risk evaluation of ground settlement was conducted by carrying out Monte Carlo simulation for the integrated model of soil strata, groundwater and ground settlement.
The uncertainties of the soil strata model were mapped in the form of median, standard deviation, skewnesss, etc. from different soil layers. From sensitivity analysis, it could be inferred that the most influential parameters on the thickness a soil layer would be the upper and lower boundary elevations of that layer. The risk areas of building damage have been mapped where the 50th and 95th percentile of the calculated ground settlement exceeded critical values. The most influential parameters on ground settlement were found varied in different places. More efforts and resources could be spent on these parameters to decrease the unacceptable risks.
It was conclude that kriging interpolation was an effective way for generating soil strata model from boreholes.
Keywords: Kriging, Monte Carlo simulation, Soil strata, Uncertainty analysis, Sensitivity analysis, Risk analysis

• author: Minyi, Pan
• supervisor: Jonas Sundell; Andreas Persson
• organization: Dept of Physical Geography and Ecosystem Science
• alternative title: 3D Soil layer model generation for ground settlement risk evaluation
• course: NGEM01 20122
• year: 2013
• type: H2 - Master's Degree (Two Years)
• subject: Earth and Environmental Sciences
• keywords: Monte Carlo simulation, kriging, soil strata, sensitivity analysis, risk analysis, uncertainty analysis, physical geography
• publication/series: Student thesis series INES
• report number: 264
• language: English
• additional info: External supervisor Jonas Sundell, Section of Built Environment Engineering; Cowi AB, Göteborg
• id: 3411964
• date added to LUP: 2013-01-29 11:49:49
• date last changed: 2013-01-29 11:49:49